OWL DL is very restrictive

I have been working on the test cases.

A current draft can be found:
http://lists.w3.org/Archives/Public/www-archive/2002Dec/att-0068/00-t

I have been flagging files and tests as to whether they are in OWL Lite, OWL
DL, or OWL Full.
*None* of the previous tests are in OWL DL or OWL Lite.

Maybe I have misunderstood what OWL DL is.

As I understand it OWL DL is the range of the mapping from the abstract
syntax.

Thus, even a test, such as:

http://www.w3.org/TR/owl-test/#proposedRDFXMLFunction-FunctionalProperty005

(proposed by Peter)

fails to be in OWL DL because it is missing triples that are always produced
in the mapping.

e.g. considering the premises:

<rdf:RDF xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#"
  xmlns:owl ="http://www.w3.org/2002/07/owl#"
  xmlns:eg ="http://www.example.org/">
    <owl:FunctionalProperty rdf:about="http://www.example.org/foo#prop" />
    <owl:Thing rdf:about="http://www.example.org/foo#object" />
</rdf:RDF>

(or as triples)
eg:foo#prop rdf:type owl:FunctionalProperty .
eg:foo#object rdf:type owl:Thing .


This is not OWL DL, not because of any substantive reason, but simply
because in the abstract syntax it is not possible syntactically to say that

eg:foo#prop rdf:type owl:FunctionalProperty .

without also saying

eg:foo#prop rdfs:domain owl:Thing .

and choosing between saying either:

eg:foo#prop rdfs:range rdfs:Literal .
eg:foo#prop rdf:type owl:DatatypeProperty .

or:

eg:foo#prop rdfs:range owl:Thing .
eg:foo#prop rdf:type owl:ObjectProperty .


These restrictions are such that, any hand written RDF/XML will fail to be
OWL Lite or OWL DL, unless immense care is taken to repeatedly say the
obvious.

I could go as far as to suggest that the restrictions will make OWL DL
unusable with RDF/XML.

All-in-all we seem to be creating a situation where two communities will
speak two varieties of OWL and any interoperation between them will be quite
unlikely.

Jeremy

Received on Tuesday, 17 December 2002 10:42:17 UTC